CN102306495B

CN102306495B - Signal handling equipment and signal processing method

Info

Publication number: CN102306495B
Application number: CN201110091514.9A
Authority: CN
Inventors: 东山惠祐; 安部素嗣
Original assignee: Sony Corp
Current assignee: Sony Corp
Priority date: 2010-04-14
Filing date: 2011-04-07
Publication date: 2015-11-25
Anticipated expiration: 2031-04-07
Also published as: CN102306495A; JP2011237753A; US20110255710A1; US9002489B2

Abstract

The present invention relates to signal handling equipment and signal processing method.Signal handling equipment comprises: absolute value element, is configured to audio signal to convert absolute value to; Typical value computing unit, is configured to the typical value of the consecutive sample values that the maximum sample value calculation block among to the value of the sample that the preparation area block of the audio signal being converted into absolute value at least uses block to comprise comprises; Average calculation unit, is configured to the fragment comprising the continuous block of predetermined quantity to be defined as frame and calculates the mean value of the typical value of the block that the maximum of the typical value of block that frame comprises and frame comprise; And detector, be configured to take noise based on maximum to the card in the ratio test frame of mean value.

Description

Signal handling equipment and signal processing method

Technical field

The present invention relates to signal handling equipment, signal processing method and program, in particular to more easily and more reliably monitoring the signal handling equipment of noise, signal processing method and program.

Background technology

Such as the integrated microphone of such as IC register is being used to gather in the equipment of sound, likely due to user's touch apparatus and generate the noise being called " touch noise " when gathering sound.

Especially, due to energy accumulative in short time period when clicking various functional switch during recording generate card take noise and card take noise reproduced with gathered sound time do not covered by other sound and stimulate the form of the extraordinary noise of ear to export.Therefore, exist detecting and reducing the requirement that the technology of noise taken by this card.

As the method for taking noise for reducing card, proposed for use high pass filter to perform filter process to signal to be processed and use the ratio of maximum to moving average come test card take noise method (for example, see the open No.7-105692 of Japanese Examined Patent application) and for use the difference in frame between minimum value and maximum method (for example, see Japan Patent No.3420831) that test card takes noise.

But, in these methods, if signal to be processed comprises corresponding high-octane part and corresponding low-energy part, then not only card can be taken noise but also music, voice (particularly consonant) etc. can be detected as card and take noise.Such as, may be that noise taken by card having other input of high energy level in certain period.

Therefore, proposed the persistence length for detecting pulse signal and determined that signal is not that card takes noise but the method for music signal (for example, see Japan Patent No.2702446) when persistence length is equal to or greater than certain length.

Summary of the invention

But for detecting in the method for persistence length, high pass filter and low pass filter are used for test card and take noise, in addition, low pass filter must have more precipitous characteristic.Therefore, amount of calculation inevitably becomes large.

Expect more easily and more reliably detection noise.

According to embodiments of the invention, provide signal handling equipment, comprising: absolute-value device, for audio signal is converted to absolute value; Typical value calculation element, at least using the typical value of consecutive sample values included in the maximum sample value calculation block among the value of sample included in block to each block of the audio signal being converted into absolute value; Average computing device, for being defined as the fragment comprising the continuous block of predetermined quantity frame and calculating the mean value of typical value of block included in the maximum of typical value of block included in frame and frame; And checkout gear, for taking noise based on maximum to the card in the ratio test frame of mean value, maximum sample value among the value of sample included in typical value calculation element determination block corresponds to the typical value of each block, signal handling equipment comprises further: past interpolated waveforms generating apparatus, and for using the card that comprises of length and audio signal, to take the noise segments of noise identical and be positioned at and generate the past interpolated waveforms that will be used for noise segments interpolation relative to the first waveform of the fragment of the audio signal on the past side of noise segments; Following interpolated waveforms generating apparatus, generates the following interpolated waveforms that will be used for noise segments interpolation for using length identical with noise segments and being positioned at relative to the second waveform of the fragment of the audio signal on the following side of the noise segments of audio signal; Interpolated waveforms generating apparatus, generates interpolated waveforms for the interpolated waveforms of using over and following interpolated waveforms by Cross fades; And alternative, reduce card for the noise segments by replacing audio signal with interpolated waveforms and take noise.

Checkout gear can determine that when the ratio of maximum to mean value is equal to or greater than predetermined threshold frame comprises card and takes noise.

Checkout gear can use the card of the maximum of frame to be processed and mean value and the maximum and average detection frame to be processed that are arranged in other frame near frame to be processed to take noise.

Signal handling equipment may further include: noise segments checkout gear, for determining when test card takes noise in frame to be processed that noise starts block corresponding to a block in following this block: block has the typical value being equal to or less than threshold value, threshold value is one of typical value of the frame be positioned at before immediately frame to be processed, and block in the past side is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed, and noise segments checkout gear is for detecting the position of a sample in following sample: first sample performs zero passage and be positioned at and start relative to noise the last sample side in the past that block comprises.

Signal handling equipment may further include: noise segments checkout gear, for determining when test card takes noise in frame to be processed that noise stops block corresponding to a block in following this block: block has the typical value being equal to or less than threshold value, threshold value is one of typical value of the frame be arranged in after immediately frame to be processed, and block is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed on following side, and noise segments checkout gear is for detecting the position of a sample in following sample: first sample performs zero passage and be positioned at and stop the original samples that comprises of block in following side relative to noise.

Past interpolated waveforms generating apparatus can reverse by and the first waveform time of implementation in the past side being in the fragment of the audio signal adjacent with noise segments identical with noise segments to length and generate interpolated waveforms.Following interpolated waveforms generating apparatus can be reversed by and the second waveform time of implementation being in the fragment of the audio signal adjacent with noise segments on following side identical with noise segments to length and generate following interpolated waveforms.

Past interpolated waveforms generating apparatus can the symbol of sample value to the reversion of the first waveform time of implementation and before and after the end sample in the past side being positioned at noise segments different from each other time interpolated waveforms is in the past generated to the symbol negate of the sample value before and after end sample side being in the past positioned at noise segments.When the symbol of the sample value of following interpolated waveforms generating apparatus to the reversion of the second waveform time of implementation and before and after the end sample being positioned at noise segments on following side is different from each other, following interpolated waveforms is generated to the symbol negate of the sample value before and after the end sample being positioned at noise segments on following side.

Signal handling equipment may further include: noise segments checkout gear, for determining when test card takes noise in frame to be processed that the position of starting position corresponding to the original samples of a block in following this block of noise taken by card: block has the typical value being equal to or less than threshold value, threshold value is corresponding to one of the typical value of the frame before being positioned at close to frame to be processed, and block in the past side is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed.

Signal handling equipment may further include: noise segments checkout gear, for determining when test card takes noise in frame to be processed that the position of final position corresponding to the last sample of a block in following this block of noise taken by card: block has the typical value being equal to or less than threshold value, threshold value is corresponding to one of the typical value of the frame after being positioned at close to frame to be processed, and block in the past side is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed.

Alternative can have predetermined length by using and is positioned at the waveform of the fragment before the noise segments of immediately audio signal and has predetermined length and the waveform being positioned at the fragment before fragment corresponding to the first waveform of immediately audio signal performs Cross fades generates adjacent interpolated waveforms, and replaces adjacent segment with adjacent interpolated waveforms.

Alternative can have predetermined length by using and is positioned at the waveform of the fragment after the noise segments of immediately audio signal and has predetermined length and the waveform being positioned at the fragment after fragment corresponding to the second waveform of immediately audio signal performs Cross fades generates adjacent interpolated waveforms, and replaces adjacent segment with adjacent interpolated waveforms.

According to another embodiment of the present invention, provide signal processing method, comprise step: audio signal is converted to absolute value; Each block of the audio signal being converted into absolute value is at least used to the typical value of consecutive sample values included in the maximum sample value calculation block among the value of sample included in block; The fragment comprising the continuous block of predetermined quantity is defined as frame and calculates the mean value of typical value of block included in the maximum of typical value of block included in frame and frame; Based on maximum, noise is taken to the card in the ratio test frame of mean value; Maximum sample value among the value determining the sample that block comprises corresponds to the typical value of each block; The card that comprises of length and audio signal is used to take identical and the first waveform be positioned at relative to the fragment of the audio signal on the past side of noise segments of the noise segments of noise and generate and will be used for comprising the past interpolated waveforms that the interpolation of the noise segments of noise taken by card; Use length and second waveform that be positioned at relative to the fragment of the audio signal following side of the noise segments of audio signal on identical with noise segments and generate the following interpolated waveforms of the interpolation that will be used for noise segments; Use interpolated waveforms and following interpolated waveforms in the past to carry out Cross fades and generate interpolated waveforms; And take noise by reducing card by the noise segments of interpolated waveforms replacement audio signal.

According to a further embodiment of the invention, provide program, be provided for computer and perform the process comprised the steps: audio signal is converted to absolute value; Each block of the audio signal being converted into absolute value is at least used to the typical value of consecutive sample values included in the maximum sample value calculation block among the value of sample included in block; The fragment comprising the continuous block of predetermined quantity is defined as frame and calculates the mean value of typical value of block included in the maximum of typical value of block included in frame and frame; And based on maximum, noise is taken to the card in the ratio test frame of mean value.

Therefore, can more easily and more reliably detection noise.

Accompanying drawing explanation

Fig. 1 is the figure of the profile instance of the signal handling equipment illustrated according to the embodiment of the present invention;

Fig. 2 is the figure of the configuration that noise detection unit is shown;

Fig. 3 is the figure of the configuration that noise reduction element is shown;

Fig. 4 illustrates that noise reduces the flow chart of process;

Fig. 5 is the figure that input signal is shown;

Fig. 6 is the figure of the typical value that block is shown;

Fig. 7 illustrates that test card takes the figure of noise;

Fig. 8 illustrates that another test card takes the figure of noise;

Fig. 9 illustrates that further test card takes the figure of noise;

Figure 10 illustrates that further test card takes the figure of noise;

Figure 11 illustrates the figure generating interpolated waveforms;

Figure 12 illustrates that another generates the figure of interpolated waveforms;

Figure 13 illustrates the figure generating interpolated waveforms further;

Figure 14 illustrates the figure further generating interpolated waveforms;

Figure 15 illustrates that noise reduces the flow chart of process;

Figure 16 illustrates the figure generating interpolated waveforms; And

Figure 17 is the block diagram of the configuration that computer is shown.

Embodiment

Hereinafter with reference to accompanying drawing, embodiments of the invention are described.

First embodiment

The configuration of signal handling equipment

Fig. 1 is the figure of the configuration of the signal handling equipment illustrated according to the embodiment of the present invention.

Such as, signal handling equipment 11 corresponds to and gathers the recording/reproducing apparatus that ambient sound also reproduces the sound gathered.To the signal of the such as voice signal that signal handling equipment 11 input uses microphone etc. to gather.Noise taken by the card that signal handling equipment 11 detects in input signal, removes card and takes noise, and takes the signal output of noise as output signal using removing card.

Signal handling equipment 11 comprises noise detection unit 21 and noise reduction element 22.Input signal is supplied to noise detection unit 21 and noise reduction element 22.

Noise detection unit 21 detect input signal comprise card take noise fragment and detect result be supplied to noise reduction element 22.Note, card is taken noise and is corresponded to the signal comprising concentrated more macro-energy (amplitude) compared to the pole short-movie section of fragment around other on the time orientation of signal.

Noise reduction element 22 suitably removes card according to the testing result that noise taken by the card provided from noise detection unit 21 and takes noise and export gained signal from input signal.

The configuration of noise detection unit

Noise detection unit 21 as shown in Fig. 2 shown in detailed configuration Fig. 1.Particularly, noise detection unit 21 comprises full-wave rectifying circuit 51, typical value determining unit 52, average calculation unit 53 and determining unit 54.

Full-wave rectifying circuit 51 converts absolute value to input signal and absolute value is supplied to typical value determining unit 52.Typical value determining unit 52 converting absolute value to and the division of signal provided from full-wave rectifying circuit 51 becomes the block corresponding with the fragment separately with predetermined length, the typical value of calculation block, and is supplied to average calculation unit 53 typical value.Such as, the maximum among the sample value of input signal that comprises of block is used as the typical value of the sample value of block.

Average calculation unit 53 uses the typical value of the block provided from typical value determining unit 52 calculate maximum and the mean value of continuous block included frame and maximum and mean value are supplied to determining unit 54.Determining unit 54 obtains the ratio of maximum to mean value of the frame provided from average calculation unit 53, whether comprises card take noise according to ratio determination frame, and the testing result that determination result takes noise as card is supplied to noise reduction element 22.

The configuration of noise reduction element

In addition, the noise reduction element 22 as shown in Fig. 3 shown in allocation plan 1.

Particularly, noise reduction element 22 comprises noise segments determining unit 81, past interpolated waveforms generation unit 82, following interpolated waveforms generation unit 83, synthesis unit 84 and replacement unit 85.In noise reduction element 22, signal is input to noise segments determining unit 81, past interpolated waveforms generation unit 82, following interpolated waveforms generation unit 83 and replacement unit 85.

The testing result that noise segments determining unit 81 takes noise according to the card provided from determining unit 54 assert that input signal comprises card and takes the fragment of noise and the result of identification is supplied to over interpolated waveforms generation unit 82, following interpolated waveforms generation unit 83 and replacement unit 85.Note, input signal can be comprised the fragment that card takes noise hereinafter and be called " noise segments ".

Past interpolated waveforms generation unit 82 uses the fragment before noise segments included in the input signal in time to generate past interpolated waveforms for noise segments interpolation according to input signal and the identification result that provides from noise segments determining unit 81, and an interpolated waveforms is in the past supplied to synthesis unit 84.

Following interpolated waveforms generation unit 83 uses the fragment after noise segments included in the input signal in time to generate following interpolated waveforms for noise segments interpolation according to input signal with from the identification result that noise segments determining unit 81 provides, and following interpolated waveforms is supplied to synthesis unit 84.

Gained interpolated waveforms by the past interpolated waveforms provided from past interpolated waveforms generation unit 82 and the following interpolated waveforms synthesis provided from following interpolated waveforms generation unit 83, and is supplied to replacement unit 85 by synthesis unit 84.Replacement unit 85 uses the identification result provided from noise segments determining unit 81 to remove card by the noise segments comprised with the interpolated waveforms replacement input signal provided from synthesis unit 84 and takes noise, and exports gained signal.

Noise reducing processes

Referring now to Fig. 4, by the noise reducing processes that description signal handling equipment 11 performs.

In step s 11, full-wave rectifying circuit 51 pairs of input signals perform full-wave rectification, that is, input signal is converted to absolute value, and income value is supplied to typical value determining unit 52.

When such as providing the input signal of the waveform shown in the top that has in Fig. 5, obtain the absolute value of the value of the sample as shown in the bottom in Fig. 5.The absolute value obtained newly is defined as the value of the sample that subjected to full-wave rectification.

Note, in Figure 5, axis of abscissas represents the time and axis of ordinates represents amplitude.In example in Figure 5, be positioned at the sample value of the sample of the immediate vicinity of input signal, i.e. amplitude (energy), compared to the remarkable protrusion of the value of sample around other.That is, amplitude significantly changes in the short-movie section of immediate vicinity, and the amplitude of fragment is greater than the amplitude of fragment around.

As mentioned above, have among the waveform of scheduled duration, the waveform that only amplitude is large in significantly short fragment is being defined as the waveform that noise taken by card.Also the noise with this waveform is called the impulsive noise or secondary (Petit) noise that stimulate ear.

In signal handling equipment 11, when wanting test card to take noise, input signal is converted to absolute value.But, because people's ear does not take noise by the Symbol recognition card of range value, so input value is converted to absolute value do not affect the detection that noise taken by card.Note, due to the remarkable change (that is, the violent increase of short time period internal power and minimizing) of amplitude, identification card takes noise to people's ear.

Referring back to the flow chart of example in Fig. 4, after input signal is converted to absolute value, typical value determining unit 52 is in step s 12 being converted into absolute value and being divided into block from the input signal that full-wave rectifying circuit 51 provides and obtaining the typical value that will be supplied to average calculation unit 53.

As illustrated in figure 6, typical value determining unit 52 is such as divided into the block corresponding with the fragment of four samples continuously arranged on the time orientation being included in input signal separately input signal.Note, in figure 6, circle represents each sample of input signal, the position representative sample value of sample on vertical direction.In example shown in Figure 6, input signal is divided into nine blocks comprising block BK1 to BK9.Typical value determining unit 52 is defined as the typical value of block the maximum among the sample value of four samples included in each block.

In step s 13, average calculation unit 53 uses the typical value of the block provided from typical value determining unit 52 to obtain frame the mean value of the typical value of included block and maximum and maximum and mean value are supplied to determining unit 54.

Such as, as shown in Figure 6, average calculation unit 53 is defined as a frame the fragment being included in continuous print nine block BK1 to BK9 on time orientation, and this frame is defined as frame to be processed.Subsequently, average calculation unit 53 obtains mean value and the maximum of the typical value of block BK1 to BK9 included in frame.

Such as, in example shown in Figure 6, maximum among the typical value due to the block included in frame of the typical value of block BK5, so the typical value of block BK5 is defined as the maximum PK of the typical value of frame.In addition, the mean value AVC of the typical value of block is greater than the mean value AVS of the value of all samples included in frame.

In step S14, determining unit 54 obtains the ratio of maximum to mean value to each frame provided from average calculation unit 53.Such as, when to represent with PK in frame to be processed included block typical value maximum and represent the mean value of typical value of block included in frame with AVC time, the ratio R T that determining unit 54 calculates maximum and mean value is as follows: RT=(PK/AVC).

In step S15, according to the ratio R T of maximum to mean value obtained, determining unit 54 determines whether frame to be processed comprises card and take noise.Particularly, when the ratio R T obtained is equal to or greater than predetermined threshold th, determine that frame to be processed comprises card and takes noise.

Such as, when threshold value th is " 3 ", in the example that maximum PK is shown in Figure 6 larger than mean value AVC three times or more.Therefore, determine that frame comprises card and takes noise.In this case, the block BK5 with maximum PK should comprise card and take noise.

In signal handling equipment 11, improve owing to using the mean value of typical value of block and the mean value of the sample value of non-input signal the detection accuracy that noise taken by card.

Assuming that as shown in the top in Fig. 7, input the signal in some samples and all samples comprised in the large sample of amplitude (sample value) with little amplitude average value.Note, in the figure 7, axis of ordinates represents input signal amplitude, and axis of abscissas represents the time.

Although the input signal shown in the top in Fig. 7 comprises the fragment that amplitude acutely changes, the amplitude near some fragments of this fragment also changes tempestuously.Therefore, can not be that noise taken by card input signal detection, that is, input signal can correspond to the normal sound of such as music.

When input signal will be processed, input signal is converted to absolute value.Thus, the input signal shown in bottom in Fig. 7 is obtained.Input signal shown in bottom in Fig. 7 comprises the sample with large amplitude by equal intervals.

Subsequently, the input signal being converted into absolute value is divided into block as shown in Figure 8, obtains maximum and the mean value of the typical value of block included in the fragment corresponding with frame.Note, in fig. 8, axis of ordinates represents the amplitude of input signal and axis of abscissas represents the time.In addition, frame comprises nine continuous block BK21 to BK29 of input signal.In this frame, obtain the maximum PK21 of typical value of block and the mean value AVC21 of the typical value of block.

The threshold value th supposing to take for card noise measuring is herein " 3 ", because the ratio (RT=(PK21/AVC21)) of maximum to mean value is less than threshold value th " 3 ", in this example so reliably determine that frame does not comprise card and takes noise.

On the other hand, the ratio (PK21/AVC21) of maximum PK21 to mean value AVS21 of the value of included in frame all samples is equal to or greater than threshold value th " 3 ".Therefore, if by this ratio is performed whether comprise to frame to be processed the determination that noise taken by card compared with threshold value th, then can determine that normal sound wavelength corresponds to card and takes noise.

As mentioned above, take noise by the maximum of the typical value using block relative to the ratio test card of the mean value of the typical value of block, reliably identify the waveform (fluctuation) of whole frame, further improve the accuracy of detection.That is, can determine whether the input signal even significantly changed in some fragments comprises card and take noise, and this input signal is likely erroneously detected as card and takes noise, such as has the audio signal of less mean value in whole amplitude more reliably.

Note, although in the above description, use the mean value of the typical value of block included in frame and maximum determination frame whether to comprise card and take noise, not only can use frame to be processed that frame to be processed and neighbouring frame thereof can also be used to make and determine.When using the multiple frame execute cards comprising frame to be processed to take the detection of noise, the detection accuracy that noise taken by card can be improved further.

Assuming that input there is the audio volume control shown in Fig. 9 signal as input signal.Note, in fig .9, axis of ordinates represents amplitude, and axis of abscissas represents the time.

The waveform of the audio signal that the sound " ka " that the audio volume control shown in Fig. 9 corresponds to the pass collection people generation obtains.The waveform of this sound started with consonant " t ", " k " or " p " with such as specified by arrow A 11 with card take noise similarly impulse form raise, after this, reduce the rank of amplitude.Subsequently, tone (pitch) waveform continues as specifiedly in arrow A 12.

Owing to generating waveform when producing sound " ka ", so waveform does not represent card take noise.But, comprise at frame to be processed rising part that arrow A 11 indicates but do not comprise in the situation of the tonal waveform part that arrow A 12 indicates, if only use a frame execute card in frame to take the detection of noise, then there will be error detection.That is, the consonant part corresponding with the start-up portion of the sound that arrow A 11 represents can be detected as card and take noise.

Therefore, when the typical value execute card of the block using some frames in frame takes the detection of noise, the accuracy of detection is further improved.Particularly, assuming that the input signal with the audio volume control shown in Fig. 9 is divided into three frames F (n) as shown in Figure 10 to F (n+2).Note, in Fig. 10, axis of ordinates represents amplitude, and axis of abscissas represents the time.In addition, the circle shown in Figure 10 represents each sample of input signal.

In example shown in Figure 10, frame F (n) comprises the rising part of audio volume control, that is, consonant part.Frame F (n+1) comprises the part between tonal waveform part and consonant part.In addition, other frame F (n+2) comprises the part of tonal waveform.Note, in the input signal, frame F (n) be relative to other frame F (n+1) and F (n+2) at front frame.

When mean value and the maximum of the typical value to each frame acquisition block, maximum PK (n) and mean value AVC (n) is obtained in frame F (n), in frame F (n+1), obtain maximum PK (n+1) and mean value AVC (n+1), in frame F (n+2), obtain maximum PK (n+2) and mean value AVC (n+2).

Herein, in frame F (n) and F (n+2), maximum PK (n) and PK (n+2) is larger to a certain extent due to consonant part and tonal waveform part.On the other hand, because frame F (n+1) does not comprise the large sample of amplitude, so maximum PK (n+1) is smaller.

In addition, because frame F (n) and F (n+1) only include the large sample of a small amount of amplitude, so mean value AVC (n) is smaller with AVC (n+1).On the other hand, in the frame F (n+2) comprising the large tonal waveform of amplitude, mean value AVC (n+2) is larger.

Present supposition frame F (n) is corresponding to frame to be processed.Such as, the maximum PK (n) that determining unit 54 obtains frame F (n) to be processed respectively to frame F (n) to each mean value AVC (n) of F (n+2) to the ratio of AVC (n+2), and each ratio compared with threshold value th.

Subsequently, in the condition meeting ((PK (n)/AVC (n) >=th), (PK (n)/AVC (n+1) >=th) and (PK (n)/AVC (n+2) >=th)), determine that frame F (n) to be processed comprises card and takes noise.Namely, when maximum PK (n) is greater than the value by frame F (n) to each mean value in the mean value of F (n+2) being multiplied by threshold value acquisition, the amplitude only with the block part that maximum PK (n) is representatively worth significantly can be protruded in continuous three frames.Therefore, in this case, determine that frame F (n) comprises card and takes noise.

In addition, in the situation meeting inequality PK (n)/AVC (n) >=th and PK (n)/AVC (n+2) < th, maximum PK (n) does not significantly protrude compared to the degree of the average amplitude of frame F (n+2) and does not correspond to card and takes noise.Therefore, in this case, determine that frame F (n) does not comprise card and takes noise.

As mentioned above, by the maximum of frame to be processed with near frame to be processed other frame mean value compared with, the detection accuracy that noise taken by card can be improved.

Note, noise can be taken to make the maximum of frame to be processed compared with the maximum of other frame near frame to be processed by another way test card.In this case, when the maximum PK (n) of such as frame F (n) to be processed predetermined value larger than maximum PK (n+1) and PK (n+2), determine that frame F (n) comprises card and takes noise.

Referring back to the flow chart shown in Fig. 4, when determine in step S15 frame do not comprise card take noise time, determining unit 54 does not comprise the determination result that card takes noise be supplied to noise segments determining unit 81 representing frame to be processed.

Subsequently, noise segments determining unit 81 exports according to the determination result instruction replacement unit 85 provided from determining unit 54 output signal representing the frame to be processed of input signal.Replacement unit 85 exports the output signal of the representative fragment corresponding with the frame to be processed of input signal according to the instruction provided from noise segments determining unit 81, and after this, process proceeds to step S21.

On the other hand, when determine in step S15 frame comprise card take noise time, determining unit 54 comprises the determination result that card takes noise be supplied to noise segments determining unit 81 representing frame to be processed, and after this, process proceeds to step S16.

Herein, representative comprises determination result that card takes noise and comprises frame to be processed and adjacent with frame to be processed with the mean value of the typical value of block, the maximum of typical value and the typical value included by frame to be processed being clipped in central frame.

In step s 16, the determination result identification that noise segments determining unit 81 uses the card provided from determining unit 54 to take noise comprises in the fragment that the frame to be processed with input signal is corresponding the noise segments that noise taken by card.

Such as, as as shown in the top of Figure 11, assuming that determining unit 54 is supplied to noise segments determining unit 81 continuously arranged three frame F (n-1) in time to the mean value AVC (n-1) to AVC (n+1) of F (n+1) and maximum PK (n-1) to PK (n+1).In addition, assuming that determining unit 54 is supplied to noise segments determining unit 81 the typical value of block included in frame F (n-1) to F (n+1).

Note, in fig. 11, axis of abscissas represents the time, and axis of ordinates represents the amplitude of input signal.In addition, frame F (n-1) be relative to other frame F (n) and F (n+1) at front frame.

In fig. 11, frame F (n-1) comprises six block BK (n-1)-1 to BK (n-1)-6.Similarly, frame F (n) comprises block BK (n)-1 to BK (n)-6, and frame F (n+1) comprises block BK (n+1)-1 to BK (n+1)-6.In addition, in frame F (n) to be processed, block BK (n)-4 has the typical value as maximum PK (n).Note, in top in fig. 11, circle represents each sample of input signal, the vertical position representative sample value of sample.

First, noise segments determining unit 81 detects the starting position that the noise segments of noise taken by the card comprising block BK (n)-4 of the typical value had as maximum PK (n), that is, the left end of noise segments in figure.In this case, noise segments determining unit 81 be used as relative to frame F (n) to be processed at front frame and the mean value AVC (n-1) being placed in the typical value of the block of the frame F (n-1) adjacent with frame F (n) to be processed as threshold value ths.

Subsequently, noise segments determining unit 81 in the past direction is detected first block that typical value is less than threshold value ths from take block BK (n)-4 at noise center as card.The block detected is defined as noise and starts block.

Assuming that in fig. 11, the typical value in the past direction being positioned at block BK (n)-3 adjacent with block BK (n)-4 is greater than threshold value ths, and the typical value of direction (left side in Figure 11) upper block BK (n)-2 adjacent with block BK (n)-3 is equal to or less than threshold value ths in the past.In this case, block BK (n)-2 is that typical value is equal to or less than first block of threshold value ths on direction in the past.Therefore, block BK (n)-2 is defined as noise and starts block.

In addition, noise segments determining unit 81 with reference to be used as the noise of input signal and start fragment corresponding to block BK (n)-2 of block to assert from the last sample in fragment (block) the sample first direction in the past performing zero passage.Subsequently, the position of the sample assert is defined as the starting position of noise segments.

Such as, specified by arrow A 41 as shown in Figure 11, assert following sample: this sample has symbol and the value that in the fragment of block BK (n)-2 corresponding to input signal, the symbol of the value of sample is contrary the most rearward, namely, the last sample of fragment, and be positioned on the most following side among the sample in the fragment that block BK (n)-2 with input signal is corresponding.

In fig. 11, the fragment that will process the input signal corresponding with block BK (n)-2 that arrow A 41 indicates is determined.Note, in fig. 11, circle represents each sample of input signal, the vertical position representative sample value of sample.Such as, corresponding with the circle on the top being arranged in figure vertical curve sample has positive sample value and the sample corresponding with the circle of the bottom being positioned at vertical curve has negative sample value.In addition, in fig. 11, horizontal direction represents the time, and especially, right direction corresponds to future directions.

Herein, in the part of the input signal indicated in arrow A 41, the sample SP11 being arranged in figure right-hand member corresponds to the last sample of the fragment corresponding with block BK (n)-2 of input signal, that is, the nearest sample in fragment.Value due to sample SP11 be on the occasion of, so there is negative value, be positioned at relative to sample SP11 past direction on and the sample be positioned near sample SP11 corresponding to the sample of starting position being positioned at noise segments.Therefore, in fig. 11, the sample SP12 being positioned at first three sample of sample SP11 is in time corresponding to the sample of starting position being positioned at noise segments.

After the starting position assert noise segments in this way, noise segments determining unit 81 test card takes the final position of the noise segments of noise, namely, the right-hand member of noise segments in figure, block BK (n)-4 that noise comprises the maximum PK (n) had as typical value taken by card.In this case, noise segments determining unit 81 is used in the mean value AVC (n+1) of typical value future directions being arranged in the included block of the frame F (n+1) adjacent with frame F (n) to be processed as threshold value the.

Noise segments determining unit 81 from take as card noise center block BK (n)-4 on future directions, detect first block that typical value is equal to or less than threshold value the, and the block detected be defined as stop block.

Assuming that in fig. 11, the typical value that future directions is positioned at block BK (n)-5 adjacent with block BK (n)-4 is greater than threshold value the, and the typical value being positioned at block BK (n)-6 adjacent with block BK (n)-5 on future directions (right side in figure) is equal to or less than threshold value the.In this case, when checking from block BK (n)-4, block BK (n)-6 is that first typical value is equal to or less than threshold value the and is positioned at the block on following side.Therefore, determine that block BK (n)-6 corresponds to noise and stops block.

In addition, noise segments determining unit 81 with reference to be used as noise in input signal and stop fragment corresponding to block BK (n)-6 of block to assert the sample first performing zero passage from the original samples of fragment (block) on future directions.Subsequently, the position of sample is defined as the final position of noise segments.

Such as, specified by arrow A 42 as shown in Figure 11, assert following sample: this sample has the symbol value contrary with the symbol of the value of original samples SP21 in the fragment of block BK (n)-6 corresponding to input signal, namely, sample the oldest in time, and this sample is positioned on the side of passing by most among the sample in the fragment that block BK (n)-6 with input signal is corresponding.

In fig. 11, in the part in the input signal that arrow A 42 indicates, the sample SP21 being positioned at left end corresponds to the original samples of the fragment of the input signal corresponding with block BK (n)-6.Due to sample SP21 have on the occasion of, so there is negative value and being positioned at the sample that the sample be on the following side relative to sample SP21, among these samples near sample SP21 is defined as being positioned at the final position of noise segments.Therefore, in fig. 11, the sample SP22 adjacent with sample SP21 is defined as the sample of the final position being positioned at noise segments.

Identification described above from starting position to the fragment of final position, that is, from sample SP12 to the fragment of sample SP22 correspond to noise segments NZ.Note, the length of noise segments NZ is specially called " interpolation length ".

As mentioned above, in signal handling equipment 11, use and frame F (n) to be processed is clipped in the mean value of central frame as threshold value, the fragment comprising typical value and be greater than the block of threshold value is defined as noise segments NZ.

Do not comprise card take noise assuming that frame F (n) to be processed is clipped in central frame, then the mean value being positioned at the typical value of the frame before and after frame F (n) represents the significantly mean value near frame F (n) in the input signal.The typical value of taking block included in the part of noise due to card may be greater than mean value, so comprise continuously arranged typical value to be greater than the fragment of the block of mean value corresponding to blocking the fragment of taking noise.Therefore, when using the mean value of the frame before and after frame F (n) to be processed as threshold value, reliably assert that the fragment of noise taken by card.

Note, can determine card take noise with make card take the length of noise value correspond to 2 power.

In this case, if from noise starting position to the fragment of noise final position (namely, from sample SP12 to the fragment of sample SP22) sample quantity correspond to 2 power, then do not change being defined as noise segments from sample SP12 to the fragment of sample SP22.

On the other hand, when from sample SP12 to when in the fragment of sample SP22, the quantity of sample does not correspond to the power of 2, being greater than among from sample SP12 to the value corresponding to the power of 2 of the quantity of sample the fragment of sample SP22, minimum value is defined as the length of noise segments.Assuming that from sample SP12 to the quantity of sample the fragment of sample SP22 be " 368 ".Not correspond to the value of the power of 2 due to " 368 ", thus be greater than " 368 " but the value " 512 " corresponding to the minimum value of the power of 2 is defined as the length of noise segments.

In addition, when the length representative of noise segments corresponds to the value of the power of 2, the starting position of noise segments is positioned at sample SP12, that is, be positioned at the end of block from noise and check the position of the sample first performing zero passage.Therefore, the final position of noise segments is positioned at length and corresponds to the power of 2 and the clearing end of the fragment from the position of sample SP12.

As mentioned above, due to the length of noise segments is defined as being equal to or greater than from sample SP12 to the quantity of sample the fragment of sample SP22 and correspond to 2 power value among minimum value, so the amount of calculation of the interpolation processing performed in last stages can be reduced.Particularly, such as, the process in the step S19 that will be described below can only be realized by multiplication and shifting function, that is, in front interpolated waveforms and the weighted calculation that performs when the Cross fades of interpolation waveform.

In addition, in the above description, start by self noise end that block and noise stop block to play the sample that first identification perform zero passage and reliably assert noise starting position and noise final position.But, this process can not be performed.In this case, such as, the original samples that noise starts block is defined as the starting position of noise segments and the last sample that noise stops block is defined as final position.

As mentioned above, searching zero crossing and the process to each onblock executing interpolation by omitting, reducing amount of calculation, having assert noise segments at once.In this case, because the final position of noise segments and starting position may not correspond to zero crossing, so slightly can generate DC component due to the interpolation of noise segments.But its unlikely acoustic mass that makes worsens.

Referring back to the flow chart shown in Fig. 4, when assert noise segments NZ, noise segments determining unit 81 is supplied to over interpolated waveforms generation unit 82, following interpolated waveforms generation unit 83 and replacement unit 85 the information of the noise segments NZ of the representative identification of the information of the final position and starting position that such as represent noise segments NZ.After this, process proceeds to step S17 from step S16.

In step S17, past interpolated waveforms generation unit 82 uses to be had interpolation length and is positioned at the information generation past interpolated waveforms of the representative noise segments NZ provided from noise segments determining unit 81 relative to sample and the use in the past of noise starting position and past interpolated waveforms is supplied to synthesis unit 84.

Such as, when input has the signal of the waveform that the arrow A 43 shown in Figure 11 indicates, past interpolated waveforms generation unit 82 extracts to be had interpolation length and is positioned at the fragment PR before the noise segments NZ of immediately input signal, and time of implementation reversion (timereversal) is to generate over interpolated waveforms PS.

Particularly, the fragment PR of input signal is adjacent with noise segments NZ on side in the past, that is, adjacent with noise segments NZ on left side in fig. 11.In addition, the length of fragment PR equals the length of noise segments NZ.Therefore, the position at the right-hand member place of fragment PR corresponds to the position of the adjacent sample of the sample SP12 that indicates with arrow A 41 on left side.In addition, owing to obtaining interpolated waveforms PS in the past by reversing to the fragment PR time of implementation of input signal, so sample adjacent with sample SP12 on left side corresponds to the sample at the left end place of interpolated waveforms PS in the past in Figure 11.On the contrary, in Figure 11, the sample at the left end place of fragment PR corresponds to the sample at the right-hand member place of interpolated waveforms PS in the past.

In step S18, following interpolated waveforms generation unit 83 uses to be had interpolation length and is positioned at relative to the sample on the following side of noise final position and uses the information of the noise segments NZ provided from noise segments determining unit 81 to generate following interpolated waveforms, and following interpolated waveforms is supplied to synthesis unit 84.

Such as, when input there is the signal of the waveform that the arrow A 43 shown in Figure 11 indicates time, following interpolated waveforms generation unit 83 extracts has interpolation length and the fragment FR be positioned at after the noise segments NZ of immediately input signal reverse to generate following interpolated waveforms FS to the fragment FR time of implementation.

Particularly, fragment FR is adjacent with noise segments NZ on following side, that is, adjacent with noise segments NZ on right side in fig. 11.In addition, the length of fragment FR is identical with the length of noise segments NZ.Therefore, in Figure 11, the position at the left end place of fragment FR corresponds to the position of the adjacent sample of sample SP22 that right side in fig. 11 indicates with arrow A 42.In addition, owing to obtaining following interpolated waveforms FS by reversing to the fragment FR time of implementation, so sample adjacent with sample SP22 on right side corresponds to the sample at the right-hand member place of following interpolated waveforms FS in Figure 11.On the contrary, in Figure 11, the sample at the right-hand member place of fragment FR corresponds to the sample at the left end place of following interpolated waveforms FS.

As mentioned above, there is interpolation length and fragment before and after the noise segments NZ being positioned at input signal generates waveform for noise segments NZ interpolation owing to using, thus after standing interpolation in input signal near noise segments NZ the power of part can unify.Thus, obtain natural waveform and there is no strange sensation.

In addition, because before and after noise segments NZ, the fragment of input signal stands time reversal, so in the past first sample of interpolated waveforms PS and the last sample of following interpolated waveforms FS correspond respectively to the sample that is positioned at before immediately noise segments and are positioned at the sample after immediately noise segments.Therefore, when using past interpolated waveforms PS and following interpolated waveforms FS to noise segments execution interpolation, the waveform of interpolation and the connection between the waveform on the border of noise segments can to become more certainly but not have strange sensation.

Referring back to the flow chart shown in Fig. 4, in step S19, synthesis unit 84 uses the past interpolated waveforms PS that provides from past interpolated waveforms generation unit 82 and performs Cross fades to generate interpolated waveforms from the following interpolated waveforms FS that following interpolated waveforms generation unit 83 provides.

Particularly, synthesis unit 84 is multiplied by the value of sample included in past interpolated waveforms PS the weight that the arrow A 44 shown in Figure 11 indicates, the value of sample included in following interpolated waveforms FS is multiplied by the weight that arrow A 45 indicates, and will passes by interpolated waveforms PS and following interpolated waveforms FS and synthesize.

In example shown in Figure 11, being " 1 " in order to be multiplied by the weight of the sample at the left end place of interpolated waveforms PS in the past, is " 0 " in order to be multiplied by the weight of the sample at the right-hand member place of interpolated waveforms PS in the past.In addition, diminish gradually to the right in fig. 11 in order to be multiplied by over the weight of sample included in interpolated waveforms PS.

On the other hand, being " 1 " in order to be multiplied by the weight of the sample at the right-hand member place of following interpolated waveforms FS in fig. 11, is " 0 " in order to be multiplied by the weight of the sample at the left end place at following interpolated waveforms FS.In addition, the weight in order to be multiplied by sample included in following interpolated waveforms FS diminishes left in fig. 11 gradually.

Synthesis unit 84 obtain the value that to be multiplied by the past interpolated waveforms PS of weight included sample be multiplied by weight and be arranged in the value of the included sample of the following interpolated waveforms FS of the sample to correspond to interpolated waveforms PS in the past and to generate interpolated waveforms HS.Such as, the value of the value being multiplied by Figure 11 of weight the sample at the right-hand member place of interpolated waveforms PS in the past and the sample at the right-hand member place of the following interpolated waveforms FS being multiplied by weight and the value of sample at the right-hand member place that is used as interpolated waveforms HS.

Referring back to the flow chart shown in Fig. 4, after generation interpolated waveforms HS, synthesis unit 84 is supplied to replacement unit 85 the interpolated waveforms HS generated.Process proceeds to step S20 from step S19.

In step S20, the noise segments NZ that replacement unit 85 uses the information of the representative noise segments NZ provided from noise segments determining unit 81 to replace input signal with the interpolated waveforms HS provided from synthesis unit 84 takes noise to make reducing card.

Such as, when input has the signal of the waveform that the arrow A 46 shown in Figure 11 indicates, replacement unit 85 is replaced noise segments NZ with interpolated waveforms HS and is taken noise to make removing card from input signal and gained signal is outputted to later stages.

Remove noise in step S20 after or when determining that in step S15 not comprising card takes noise, process proceeds to the step S21 that signal handling equipment 11 determines whether to want termination.Such as, when to all fragments of input signal perform card take removing of noise time, determine to want termination.

When determining in the step s 21 not want termination, the processing returns to step S11, again performing aforesaid operations.That is, next frame is defined as frame to be processed, this frame execute card is taken the detection of noise and removed.

On the other hand, when determining in the step s 21 to want termination, stopping noise and reducing process.

As mentioned above, signal handling equipment 11 is divided into multiple block input signal, obtains the typical value of block, and uses the maximum of typical value of block included in frame and the ratio test card of mean value to take noise.Subsequently, signal handling equipment 11 assert that noise segments taken by the card of input signal, uses length the same with noise segments and the fragment be positioned at before and after noise segments generates interpolated waveforms, and removes card and take noise.

Thus, comprise the ratio of maximum to mean value of the typical value of the frame of block owing to calculating typical value and acquisition to each block, thus easily when amount of calculation reduces more reliably test card take noise.Therefore, can reliably remove card from input signal and take noise, in acoustics sense, obtain natural sound and there is no strange sensation.

Note, particularly, when generating past interpolated waveforms or following interpolated waveforms, if the symbol being located at the sample before or after the starting position of noise segments or the sample at final position place is different from each other, then to the symbol negate of the value of sample included in the sample group of the fragment of the input signal for interpolation.

Particularly, assuming that as shown in the top of Figure 12, sample SP41 corresponds to the peak value that noise taken by card, and sample SP42 corresponds to the starting position of noise segments.

Note, in fig. 12, circle represents each sample of input signal, the vertical position representative sample value of sample.Such as, corresponding with the circle of the upside being arranged in figure vertical curve sample representation has the sample representation corresponding with the circle of the downside being positioned at vertical curve on the occasion of the sample as sample value and has the sample of negative value as sample value.In addition, in fig. 12, horizontal direction represents the time, and especially, right direction represents future directions.

In the input signal shown in the upside of Figure 12, correspond to the noise segments that will convert interpolated waveforms to relative to the part on the right side of sample SP42.In addition, use to be included in left side to be positioned at the sample SP43 adjacent with the sample SP42 of starting position corresponding to noise segments and to be positioned at and generates relative to the sample in the past left side of figure (that is, on) of noise segments the past interpolated waveforms generated for interpolated waveforms.

In this case, whether past interpolated waveforms generation unit 82 to determine to be positioned in time the symbol of sample SP43 and SP44 before and after sample SP42 mutually the same and generate interpolated waveforms in the past.Such as, in example shown in Figure 12, symbol sample SP42 being clipped in the value of central sample SP43 and SP44 is different from each other.

Therefore, the rectangle K11 that past interpolated waveforms generation unit 82 extracts example in the core of figure around the part of input signal, that is, there is interpolation length (noise segments length) and comprise the fragment of the sample SP43 at right-hand member place in figure and this fragment time of implementation is reversed.In addition, past interpolated waveforms generation unit 82 by by rectangle K11 around input signal the part time of implementation reversion and obtain waveform sample value symbol negate with obtains pass by interpolated waveforms.Thus, as shown in the bottom of Figure 12, obtain rectangle K12 around past interpolated waveforms.

In the bottom of Figure 12, replace the noise segments of input signal with the past interpolated waveforms obtained and on the right side of the part corresponding with rectangle K11, arrange obtained past interpolated waveforms.Such as, by the symbol negate being arranged in the sample SP43 at the right-hand member place of rectangle K11 for the interpolated waveforms of generating over is obtained rectangle K12 around the value of sample at figure left end place of past interpolated waveforms.

Thus, when the symbol of the sample before and after the sample SP42 at starting position place being positioned at noise segments is different from each other, when generating in the past interpolated waveforms to the fragment of the input signal for the interpolated waveforms of generating in included by the symbol negate of sample.Therefore, when shown in the bottom in such as Figure 12 with when interpolated waveforms replaces the noise segments of input signal in the past, obtain the smooth boundary part at the starting position place in noise segments, that is, obtain input signal and the smooth connection part in the past between interpolated waveforms.As a result, when the interpolated waveforms by using past interpolated waveforms and following interpolated waveforms to perform Cross fades and obtain is arranged in noise segments, acquisition has the signal of nature waveform and does not have strange sensation.

On the other hand, as shown in the top of Figure 13, when the symbol of the value of the sample before and after the sample at starting position place being positioned at noise segments is mutually the same, when generating past interpolated waveforms not to the symbol negate of sample value.

Note, in fig. 13, also as the situation of Figure 12, circle represents each sample of input signal.

In the example shown in the top of Figure 13, the sample SP61 of input signal corresponds to the peak value that noise taken by card, and sample SP2 corresponds to the starting position of noise segments.In addition, in the input signal, correspond to noise segments relative to the part on the right side of sample SP62, replace this fragment with interpolated waveforms.In addition, to use on the left side being positioned at relative to noise segments and the sample comprising the sample SP63 that be positioned at left side on adjacent with sample SP62 generates past interpolated waveforms for generating interpolated waveforms.

At this, whether the symbol that past interpolated waveforms generation unit 82 determines to be positioned in time respectively the value of sample SP63 and SP64 before and after sample SP62 is mutually the same.Such as, in example shown in Figure 13, symbol sample SP62 being clipped in the value of central sample SP63 and SP64 is mutually the same.

Therefore, past interpolated waveforms generation unit 82 extract rectangle K31 around the part of input signal, that is, there is interpolation length and comprise the fragment of its right-hand member place sample SP63 as shown in core in figure and this fragment time of implementation is reversed to obtain interpolated waveforms in the past.Thus, as shown in the bottom of Figure 13, obtain rectangle K32 around waveform, namely pass by interpolated waveforms.

In the bottom of Figure 13, replace the noise segments of input signal with the past interpolated waveforms obtained and on the right side of the part corresponding with rectangle K31, arrange the past interpolated waveforms of acquisition.Such as, in figure rectangle K32 around the value of sample at left end place of past interpolated waveforms identical with the value being positioned at the sample SP63 at the right-hand member place of rectangle K31 for the interpolated waveforms of generating over.

As mentioned above, when the symbol of the value of the sample before and after the sample SP62 at starting position place being positioned at noise segments is mutually the same, not to the symbol negate of the value of sample included in the fragment of the input signal for the interpolated waveforms of generating over.Therefore, as shown in the bottom of Figure 13, when replacing the noise segments of input signal with past interpolated waveforms, obtain the smooth boundary part of the starting position of noise segments, that is, obtain input signal and the smooth connection part in the past between interpolated waveforms.As a result, when the interpolated waveforms by performing Cross fades to past interpolated waveforms and following interpolated waveforms and obtain is arranged in noise segments, acquisition has the signal of nature waveform and does not have strange sensation.

Note, as the situation of past interpolated waveforms, in the situation generating following interpolated waveforms, when the symbol of the value being positioned at the sample before and after noise final position is different from each other, to the symbol negate of the value of the sample for following interpolated waveforms.

In addition, in the above description, the maximum of the value of sample included in block is defined as the typical value of block.But, can determine to represent block by using the calculating of the value meeting sample included in the block of predetermined condition.Such as, typical value can be obtained by performing weighted sum to the value of all samples included in block.Alternatively, the sample of predetermined quantity can be selected by the descending of sample value, and the mean value of sample value can be defined as typical value.

Second embodiment

In the above description, replace the Related Computational Methods along with a large amount of calculating and cost, described and effectively reduced for realizing the method that noise taken by card.Amount of calculation is reduced by the waveform replaced in noise segments with interpolated waveforms.But, in this method, when reproducing the output signal obtained, the sound of (discontinuous) waveform that do not continue corresponding to output signal may be obtained near the end of the noise segments of having replaced with interpolated waveforms.

Particularly, assuming that the signal that the arrow A 61 shown in the top in Figure 14 indicates is input to signal handling equipment 11, and the fragment NZ31 of input signal is detected as noise segments (hereinafter, being called " noise segments NZ31 ").

Note, in fig. 14, axis of abscissas represents the time, and axis of ordinates represents the amplitude of input signal.In addition, in fig. 14, circle represents each sample of input signal, the vertical position representative sample value of sample.Especially, corresponding with the circle of the upside being arranged in figure vertical curve sample has and has negative value as sample value on the occasion of the sample corresponding with the circle of the downside being arranged in figure vertical curve as sample value.

Specified by arrow A 61, as detection noise fragment NZ31 in the input signal, the noise of example reduces in process in the diagram, and on time orientation, reversion has interpolation length and is positioned at fragment PR21 before the noise segments NZ31 of immediately input signal to make as specified in arrow A 62 the interpolated waveforms of generating over.Similarly, on time orientation, reversion has interpolation length and is positioned at the following interpolated waveforms that fragment FR21 after the noise segments NZ31 of immediately input signal indicates to make to generate arrow A 63.

Subsequently, as specified in arrow A 64, take noise with the noise segments NZ31 of the interpolated waveforms HS21 replacement input signal by using interpolated waveforms and following interpolated waveforms in the past to perform Cross fades and obtain to make removing card.

In this noise removal method, generate final interpolated waveforms HS21 by performing weighting according to the distance apart from noise segments NZ31, so reduce the not nature of waveform in noise segments NZ31 owing to using interpolated waveforms and following interpolated waveforms in the past.In addition, in this method, due to do not continue (discontinuity) of the final position and starting position place sample value that avoid noise segments NZ31 in principle, so obvious strange sensation and abnormal sound can not be generated.

But when the part before and after the noise segments NZ31 of input signal comprises frequency low waveform, obviously occur aliasing waveform in the part before and after the noise segments NZ31 of output signal, aliased portion has high fdrequency component.Therefore, when reproducing output signals, obtain the corresponding abnormal sound that do not continue with the waveform of output signal as a result.

In example shown in Figure 14, the fragment E11 near the starting position of the noise segments NZ31 of the input signal that the upper arrow A61 being positioned at figure indicates has the waveform similar with the sine wave of low frequency.But in the output signal that arrow A 64 represents, fragment E12 corresponding with fragment E11 in position comprises the signal with high fdrequency component waveform, therefore, can obtain unsuitable sound.

Similarly, the fragment E13 be positioned near noise segments final position has the waveform comprising high fdrequency component.This is because when card will be removed take noise, to consider in noise segments starting position and noise segments final position continue among only continuing of sample value is taken into account.

Noise reducing processes

Therefore, noise can be performed and reduce process with the waveform making the interpolation part of acquisition output signal more level and smooth.Hereinafter, with reference to Figure 15 and 16, the noise described in this situation is reduced process.

Figure 15 is the flow chart illustrating the noise reduction process that signal handling equipment 11 performs.Note, the noise of example to reduce in process the detection noise fragment in the input signal from step S51 to step S56 in fig .15, and these process are the same with the process in the step S11 to S16 of example in Fig. 4.Therefore, omit it to describe.

In step S57, past interpolated waveforms generation unit 82 uses the use of the information of the representative noise segments provided from noise segments determining unit 81 to have the generating past interpolated waveforms at front sample and a past interpolated waveforms is supplied to synthesis unit 84 relative to noise starting position of interpolation length.

Such as, when input there is the signal of the waveform that the arrow A 81 shown in Figure 16 indicates time, before past interpolated waveforms generation unit 82 extraction is positioned at the noise segments NZ41 of immediately input signal and the fragment PR31 with interpolation length as past interpolated waveforms.

Note, in figure 16, axis of abscissas represents the time, and axis of ordinates represents the amplitude of input signal.In addition, the circle shown in Figure 16 represents each sample of input signal, the vertical position representative sample value of sample.Especially, corresponding with the circle on the top being arranged in figure vertical curve sample has and has negative value as sample value on the occasion of the sample corresponding with the circle of the bottom being arranged in figure vertical curve as sample value.

In example shown in Figure 16, on the fragment PR31 corresponding with past interpolated waveforms left side in the drawings, (that is, on past side) is adjacent with noise segments NZ41, and length is identical with the length of noise segments NZ41.

In step S58, the following side that following interpolated waveforms generation unit 83 uses the information of the representative noise segments provided from noise segments determining unit 81 to use to be positioned at relative to noise final position and the sample with interpolation length generate following interpolated waveforms and following interpolated waveforms be supplied to synthesis unit 84.

Such as, when input there is the signal of the waveform that the arrow A 81 shown in Figure 16 indicates time, after following interpolated waveforms generation unit 83 extracts and is positioned at immediately noise segments NZ41 and the fragment FR31 with interpolation length as following interpolated waveforms.

As mentioned above, noise shown in Figure 15 reduces in process, and when generating past interpolated waveforms and following interpolated waveforms, the sample with interpolation length extracted is without undergoing time reversal.In addition, corresponding with past interpolated waveforms fragment PR31 and the fragment FR31 corresponding with following interpolated waveforms can not be adjacent with noise segments NZ41.

In step S59, synthesis unit 84 uses the past interpolated waveforms that provides from past interpolated waveforms generation unit 82 and performs Cross fades to generate interpolated waveforms from the following interpolated waveforms that following interpolated waveforms generation unit 83 provides.

In step S59, perform the process the same with the step S19 in Fig. 4.Particularly, obtain the in the past sample value of interpolated waveforms and the sample value of following interpolated waveforms and, the value obtained is defined as the value of sample included in interpolated waveforms.

Such as, the weight in order to the sample of the interpolated waveforms of being multiplied by over diminishes gradually towards following side, and past direction is " 1 " in the weight of front sample, and future directions is " 0 " in the weight of rear sample.On the contrary, become large in order to the weight being multiplied by the sample of following interpolated waveforms gradually towards following side, past direction is " 0 " in the weight of front sample, and future directions is " 1 " in the weight of rear sample.

Generate interpolated waveforms at synthesis unit 84 and after interpolated waveforms is supplied to replacement unit 85, processes and proceed to step S60 from step S59.

In step S60, the noise segments that replacement unit 85 uses the information of the representative noise segments provided from noise segments determining unit 81 to replace input signal with the interpolated waveforms provided from synthesis unit 84 takes noise to make the card reducing input signal.

Such as, when inputting the signal that the arrow A 82 shown in Figure 16 indicates, replacement unit 85 is replaced noise segments NZ41 with interpolated waveforms HS31 and is taken noise to make removing card from input signal.

As mentioned above, replacing simply in the state of noise segments NZ41 with interpolated waveforms HS31, be arranged in do not continue (jump of sample value) that waveform obviously appears in the border fragment PS11 near noise starting position and the border fragment FS11 be positioned near noise final position.Note, border fragment PS11 comprises noise starting position, and border fragment FS11 comprises noise final position.

Therefore, replacement unit 85 is with being replaced the waveform near the fragment PS11 of border and the waveform near the fragment FS11 of border by the newly-generated waveform of Cross fades with the generation preventing signal output waveform from not continuing.

Particularly, in step S61, replacement unit 85 perform with the fragment by using interpolated waveforms to perform to replace the noise starting position of the input signal (that is, the input signal obtained by the process performed in step S60) obtained adjacent in included by the replacement of waveform of input signal.

Particularly, specified by arrow A 83 as shown in Figure 16, replacement unit 85 is determined as predetermined short-movie section and fragment BP11 adjacent with the noise starting position of input signal on side in the past.That is, fragment BP11 is positioned at immediately before noise segments NZ41.

Next, replacement unit 85 determines and be in time positioned at the fragment BP11 of input signal before the fragment MP11 in (past) identical with fragment BP11 as intended fragment, length.In example shown in Figure 16, before fragment MP11 is positioned at immediately corresponding with past interpolated waveforms fragment PR31.

Subsequently, replacement unit 85 is used the waveform of the waveform of the fragment BP11 of input signal and the fragment MP11 of input signal to perform Cross fades and replace fragment BP11 to make to avoid not continuing of waveform with the waveform HP11 obtained by Cross fades as indicated in arrow A 84.

Such as, when performing Cross fades, diminish gradually in order to the weight being multiplied by the sample that fragment BP11 comprises towards following side, and in the past side is " 1 " in the weight of front sample, following side is " 0 " in the weight of rear sample.On the contrary, become large in order to the weight being multiplied by the sample that fragment MP11 comprises gradually towards following side, and in the past side is " 0 " in the weight of front sample, following side is " 1 " in the weight of rear sample.

Therefore, near the fragment BP11 of the input signal replaced with waveform HP11, the waveform near the final position of fragment MP11 is smoothly routed to the waveform near the starting position of fragment PR31.Therefore, not continuing of waveform is avoided.As a result, in acoustics sense, obtain natural sound and there is no strange sensation.

Particularly, when generating interpolated waveforms HS31, in order to be multiplied by that the weight of the sample at the left end place of fragment PR31 in figure is " 1 ", weight in order to the sample at the left end place of being multiplied by fragment FP31 in figure is " 0 ".Therefore, in figure, the sample at the left end place of interpolated waveforms HS31 is identical with the sample at the left end place of fragment PR31.

On the other hand, when generating waveform HP11, in order to be multiplied by that the weight of the sample at the right-hand member place of fragment MP11 in figure is " 1 ", weight in order to the sample at the right-hand member place of being multiplied by fragment BP11 in figure is " 0 ".Therefore, in figure, the sample at the right-hand member place of waveform HP11 is the same with the sample at the right-hand member place of fragment MP11.

When the waveform HP11 obtained as mentioned above being arranged in immediately before interpolated waveforms HS31, in boundary member between waveform HP11 and interpolated waveforms HS31, the sample at the left end place of fragment PR31 adjacent one another are in original input signal is become adjacent one another are with the sample permutations at the right-hand member place of fragment MP11.That is, owing to replacing the fragment BP11 of input signal with waveform HP11, so obtain nature and level and smooth waveform near the starting position of noise segments NZ41.

Referring back to the flow chart shown in Figure 15, in step S62, replacement unit 85 performs replacement to the waveform in the fragment adjacent with the noise final position of the input signal obtained by the process in step S61.

Particularly, specified by arrow A 83 as shown in Figure 16, replacement unit 85 is defined as fragment BF11 short-movie section adjacent with the noise final position in input signal on following side.In example shown in Figure 16, fragment BF11 is positioned at immediately after noise segments NZ41.

Next, replacement unit 85 is identical with fragment BF11 and the intended fragment be positioned in time after the fragment BF11 of input signal is defined as fragment MF11 length.In example shown in Figure 16, after fragment MF11 is positioned at immediately corresponding with following interpolated waveforms fragment FR31.

Subsequently, the waveform of replacement unit 85 couples of fragment BF11 and the waveform of fragment MF11 are performed Cross fades and replace the fragment BF11 of input signal to make to avoid not continuing of waveform with the waveform HF11 obtained by Cross fades as specifiedly in arrow A 84.

Such as, when performing Cross fades, become large in order to the weight being multiplied by the sample that fragment BF11 comprises gradually towards following side, and in the past side is " 0 " in the weight of front sample, following side is " 1 " in the weight of rear sample.On the contrary, diminish gradually in order to the weight being multiplied by the sample of fragment MF11 towards following side, and in the past side is " 1 " in the weight of front sample, following side is " 0 " in the weight of rear sample.

Therefore, near the fragment BF11 of the input signal replaced with waveform HF11, as the situation of fragment BP11, the waveform smooth connection each other near the waveform near the starting position of fragment MF11 and the final position of fragment FR31.As a result, avoid not continuing of waveform, in acoustics sense, obtain the natural sound corresponding with output signal and there is no strange sensation.

Replacement unit 85 outputs to follow-up phase as output signal using the input signal obtained by above-mentioned process.

Referring back to the flow chart shown in Figure 15, in step S62, perform the replacement of waveform or after determining that in step S55 not comprising card takes noise, process proceeds to step S63.

In step S63, signal handling equipment 11 determines whether to want termination.When to all fragments of input signal perform card take removing of noise time, such as, determine to want termination.

When determining not want termination in step S63, the processing returns to step S51, again performing above-mentioned process.On the other hand, when determining to want termination in step S63, stopping noise and reducing process.

As mentioned above, signal handling equipment 11 interpolated waveforms replaces the noise segments of input signal, use the fragment adjacent with the fragment for generating interpolated waveforms and the fragment newly-generated waveform adjacent with noise segments, and after this, replace the fragment adjacent with noise segments with newly-generated waveform.Thus, obtain the connection of interpolated waveforms to make to prevent from generating not continuing of waveform, and in acoustics sense, obtain natural sound and there is no strange sensation.

When using the noise of example in Figure 15 to reduce process, slightly add amount of calculation compared to the situation of Fig. 4.But, process is reduced, owing to carrying out interpolation when maintaining not the continuing of waveform to noise segments, and in addition according to the noise of example in Figure 15, interpolation is carried out to the boundary member of noise segments, so more naturally realize the reduction of noise and there is no strange sensation.

Note, although the fragment BP11 adjacent with the noise segments NZ41 shown in Figure 16 and BF11 can have any length, as long as length is no more than the length of noise segments NZ41, length should be short in as far as possible to make not reproduce strange sound.In addition, fragment BP11 and BF11 can have different length.

Above-mentioned a series of process can be performed by hardware or software.When being performed a series of process by software, the program that software is comprised is installed to can performs by installing various program the computer combined the general purpose personal computer of various function or specialized hardware from program recorded medium.

Figure 17 is the block diagram illustrating the hardware configuration being performed the computer of above-mentioned a series of process by program.

In a computer, CPU (CPU) 301, ROM (read-only memory) 302 and RAM (random access storage device) 303 are connected with each other by bus 304.

Input/output interface 305 is also connected to bus 304.For input/output interface 305, input unit 306 comprises keyboard, mouse and microphone, output unit 307 comprises display and loud speaker, record cell 308 comprises hard disk or nonvolatile memory, communication unit 309 comprises network interface, and drives the driver 310 of removable media 311 of such as disk, CD, magneto optical disk or semiconductor memory.

In the computer of as mentioned above configuration, when CPU301 by input/output interface 305 and bus 304 program of record in record cell 308 is loaded into RAM303 and executive program time, perform above-mentioned a series of process.

The computer program that (CPU301) performs is provided by being recorded in the encapsulation medium as such as disk (comprising floppy disk), CD (CD-ROM (compact disk read-only memory), DVD (digital versatile disc) etc.), magneto optical disk or semiconductor memory or by the wired or wireless transmission medium of such as local area network (LAN), the Internet or digital satellite broadcasting.

Program can be arranged in record cell 308 by input/output interface 305 by removable media 311 is inserted in driver 310.In addition, can be received by communication unit 309 by wired or wireless transmission medium and in record cell 308 installation procedure.Alternatively, program can be pre-installed in ROM302 or record cell 308.

Note, the program that will be able to perform with time Series Processing computer according to the order described in this specification, alternatively, can opportunity when caller or handling procedure concurrently.

Note, embodiments of the invention are not limited to above embodiment, can make various amendment without departing from the present invention.

The application comprises and submits to the Japanese Priority Patent Application JP2010-092817 of Japan Office and 2010 to submit the theme that disclosed in the Japanese Priority Patent Application JP2010-175335 of Japan Office, theme is relevant to 4, on Augusts on April 14th, 2010, and its full content is incorporated to herein through quoting.

It will be appreciated by those skilled in the art that needs and other factors according to design, can make various amendment, combination, sub-portfolio and conversion, as long as they are in the scope of claims or its equivalent.

Claims

1. a signal handling equipment, comprising:

Absolute-value device, for converting absolute value to audio signal;

Typical value calculation element, the typical value of the consecutive sample values that the maximum sample value calculation block at least using each block of the audio signal being converted into absolute value among the value of sample included in block comprises;

Average computing device, for being defined as frame the fragment comprising the continuous block of predetermined quantity and calculating the mean value of typical value of the block that the maximum of typical value of the block that described frame comprises and described frame comprise; And

Checkout gear, for taking noise based on described maximum to the card in frame described in the ratio test of described mean value,

Wherein, the maximum sample value among the value of the sample that described typical value calculation element determination block comprises corresponds to the typical value of each block, and

Described signal handling equipment comprises further:

Past interpolated waveforms generating apparatus, takes identical and the first waveform be positioned at relative to the fragment of the described audio signal on the past side of described noise segments of the noise segments of noise and generates and will be used for comprising for using the card that comprises of length and described audio signal the past interpolated waveforms that the interpolation of the described noise segments of noise taken by card;

Following interpolated waveforms generating apparatus, for using length identical with described noise segments and the second waveform be positioned at relative to the fragment of the described audio signal on the following side of the described noise segments of described audio signal and generate the following interpolated waveforms of the interpolation that will be used for described noise segments;

Interpolated waveforms generating apparatus, for generating interpolated waveforms by using described interpolated waveforms and described following interpolated waveforms in the past to carry out Cross fades; And

Alternative, reduces described card for the described noise segments by replacing described audio signal with described interpolated waveforms and takes noise.

2. signal handling equipment as claimed in claim 1,

Wherein, described checkout gear determines that when the ratio of described maximum to described mean value is equal to or greater than predetermined threshold described frame comprises described card and takes noise.

3. signal handling equipment as claimed in claim 1,

Wherein, described checkout gear uses the card of the described maximum of frame to be processed and described mean value and the maximum and average detection frame to be processed that are arranged in other frame near frame to be processed to take noise.

4. signal handling equipment as claimed in claim 1, comprises further:

Noise segments checkout gear, for determining that noise starts block corresponding to a block in following block when detecting in frame to be processed when noise taken by described card: block has the typical value being equal to or less than threshold value, described threshold value is one of typical value of the frame be positioned at before immediately frame to be processed, and block in the past side is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed, and described noise segments checkout gear is for detecting the position of a sample in following sample: first sample performs zero passage and be positioned at and start relative to described noise the last sample side in the past that block comprises.

5. signal handling equipment as claimed in claim 1, comprises further:

Noise segments checkout gear, for determining that noise stops block corresponding to a block in following block when detecting in frame to be processed when noise taken by described card: block has the typical value being equal to or less than threshold value, described threshold value is corresponding to one of the typical value of the frame be positioned at after immediately frame to be processed, and block is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed on following side, and described noise segments checkout gear is for detecting the position of a sample in following sample: first sample performs zero passage and be positioned at and stop original samples that block comprises in following side relative to described noise.

6. signal handling equipment as claimed in claim 1,

Wherein, described in the past interpolated waveforms generating apparatus reverse by and the described first waveform time of implementation in the past side being positioned at the fragment of the described audio signal adjacent with noise segments identical with noise segments to length and generate described pass by interpolated waveforms, and

Described following interpolated waveforms generating apparatus is reversed by and the described second waveform time of implementation being positioned at the fragment of the described audio signal adjacent with noise segments on following side identical with noise segments to length and is generated described following interpolated waveforms.

7. signal handling equipment as claimed in claim 6,

Wherein, when described in the past interpolated waveforms generating apparatus is different from each other by the symbol of the sample value before and after the end sample that to described first waveform time of implementation reversion and in the past side is positioned at noise segments, described interpolated waveforms is in the past generated to the symbol negate of the sample value before and after end sample side being in the past positioned at noise segments, and

When the symbol of the sample value of described following interpolated waveforms generating apparatus to described second waveform time of implementation reversion and before and after the end sample being positioned at noise segments on following side is different from each other, described following interpolated waveforms is generated to the symbol negate of the sample value before and after the end sample being positioned at noise segments on following side.

8. signal handling equipment as claimed in claim 1, comprises further:

Noise segments checkout gear, for determining when noise taken by described card that the starting position that noise taken by described card corresponds to the position of the original samples of one of following block when detecting in frame to be processed: block have be equal to or less than be positioned at close to frame to be processed before the typical value of threshold value corresponding to one of the typical value of frame, and block in the past side is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed.

9. signal handling equipment as claimed in claim 1, comprises further:

Noise segments checkout gear, for when detecting the position determining the last sample that the final position that noise taken by described card one of to correspond in following block when noise taken by described card in frame to be processed: block have be equal to or less than be positioned at close to frame to be processed after the typical value of threshold value corresponding to one of the typical value of frame, and block in the past side is arranged in the proximal most position of one of the block relative to the maximum typical value with frame to be processed.

10. signal handling equipment as claimed in claim 1,

Wherein, described alternative has predetermined length by using and is positioned at the waveform of the fragment before the noise segments of immediately described audio signal and has predetermined length and the waveform being positioned at the fragment before immediately corresponding with described first waveform of described audio signal fragment performs Cross fades and generates adjacent interpolated waveforms, and by described adjacent interpolated waveforms replacement adjacent segment.

11. signal handling equipments as claimed in claim 1,

Wherein, described alternative has predetermined length by using and is positioned at the waveform of the fragment after the noise segments of immediately described audio signal and has predetermined length and the waveform being positioned at the fragment after immediately corresponding with described second waveform of described audio signal fragment performs Cross fades and generates adjacent interpolated waveforms, and by described adjacent interpolated waveforms replacement adjacent segment.

12. 1 kinds of signal processing methods, comprise step:

Audio signal is converted to absolute value;

At least use the maximum sample value among the value of sample included in block to carry out the typical value of consecutive sample values included in calculation block to each block of the audio signal being converted into absolute value;

The fragment comprising the continuous block of predetermined quantity is defined as frame and calculates in described frame the mean value of the typical value of block included in the maximum of the typical value of included block and described frame;

Based on described maximum, noise is taken to the card in frame described in the ratio test of described mean value;

Maximum sample value among the value determining the sample that block comprises corresponds to the typical value of each block;

The card that comprises of length and described audio signal is used to take identical and the first waveform be positioned at relative to the fragment of the described audio signal on the past side of described noise segments of the noise segments of noise and generate and will be used for comprising the past interpolated waveforms that the interpolation of the described noise segments of noise taken by card;

Use length and second waveform that be positioned at relative to the fragment of the described audio signal following side of the described noise segments of described audio signal on identical with described noise segments and generate the following interpolated waveforms of the interpolation that will be used for described noise segments;

Use described interpolated waveforms and described following interpolated waveforms in the past to carry out Cross fades and generate interpolated waveforms; And

Reduce described card by the described noise segments of replacing described audio signal with described interpolated waveforms and take noise.

13. 1 kinds of signal handling equipments, comprising:

Absolute value element, is configured to audio signal to convert absolute value to;

Typical value computing unit, is configured at least use the maximum sample value among the value of sample included in block to carry out the typical value of consecutive sample values included in calculation block to each block of the described audio signal being converted into absolute value;

Average calculation unit, is configured to the fragment comprising the continuous block of predetermined quantity to be defined as frame and calculates in described frame the mean value of the typical value of block included in the maximum of the typical value of included block and described frame; And

Detector, noise taken by the card being configured to detect in described frame based on the ratio of described maximum to described mean value,

Wherein, the maximum sample value among the value of the sample that described typical value computing unit determination block comprises corresponds to the typical value of each block, and

Described signal handling equipment comprises further:

Past interpolated waveforms generation unit, is configured to use the card that comprises of length and described audio signal take identical and the first waveform be positioned at relative to the fragment of the described audio signal on the past side of described noise segments of the noise segments of noise and generate and will be used for comprising the past interpolated waveforms that the interpolation of the described noise segments of noise taken by card;

Following interpolated waveforms generation unit, is configured to use length identical with described noise segments and the second waveform be positioned at relative to the fragment of the described audio signal on the following side of the described noise segments of described audio signal and generate the following interpolated waveforms of the interpolation that will be used for described noise segments;

Interpolated waveforms generation unit, is configured to generate interpolated waveforms by using described interpolated waveforms and described following interpolated waveforms in the past to carry out Cross fades; And

Replacement unit, the described noise segments be configured to by replacing described audio signal with described interpolated waveforms reduces described card and takes noise.